TMCnet News
AMBAC FINANCIAL GROUP INC - 10-Q - Management's Discussion and Analysis of Financial Condition and Results of Operations.(Edgar Glimpses Via Acquire Media NewsEdge) Cautionary Statement Pursuant to the Private Securities Litigation Reform Act of 1995 In this Quarterly Report, we have included statements that may constitute "forward-looking statements" within the meaning of the safe harbor provisions of the Private Securities Litigation Reform Act of 1995. Words such as "estimate," "project," "plan," "believe," "anticipate," "intend," "planned," "potential" and similar expressions, or future or conditional verbs such as "will," "should," "would," "could," and "may", or the negative of those expressions or verbs, identify forward-looking statements. We caution readers that these statements are not guarantees of future performance. Forward-looking statements are not historical facts but instead represent only our beliefs regarding future events, which, may by their nature be inherently uncertain and some of which may be outside our control. These statements may relate to plans and objectives with respect to the future, among other things which may change. We are alerting you to the possibility that our actual results may differ, possibly materially, from the expected objectives or anticipated results that may be suggested, expressed or implied by these forward-looking statements. Important factors that could cause our results to differ, possibly materially, from those indicated in the forward-looking statements include, among others, those discussed under "Risk Factors" in Part I, Item 1A of the 2011 Annual Report on Form 10-K and Part II, Item 1A of this Form 10-Q. Any or all of management's forward-looking statements here or in other publications may turn out to be incorrect and are based on Ambac Financial Group, Inc. ("Ambac" or the "Company") management's current belief or opinions. Ambac's actual results may vary materially, and there are no guarantees about the performance of Ambac's securities. Among events, risks, uncertainties or factors that could cause actual results to differ materially are: (1) failure to consummate a plan of reorganization under Chapter 11, which may lead to the commencement of liquidation proceedings pursuant to Chapter 7; (2) the impact of the bankruptcy proceeding on the holders of Ambac securities; (3) failure to satisfactorily resolve our dispute with the United States Internal Revenue Service; (4) the unlikelihood that Ambac Assurance Corporation ("Ambac Assurance") will pay dividends to Ambac in the foreseeable future; (5) adverse events arising from the Segregated Account Rehabilitation Proceedings, including the failure of the injunctions issued by the Wisconsin Rehabilitation Court to protect the Segregated Account and Ambac Assurance from certain adverse actions; (6) litigation arising from the Segregated Account Rehabilitation Proceedings; (7) decisions made by the Rehabilitator for the benefit of policyholders may result in material adverse consequences for Ambac's securityholders; (8) potential of a full rehabilitation proceeding against Ambac Assurance or material changes to the Segregated Account Rehabilitation Plan, with resulting adverse impacts; (9) inadequacy of reserves established for losses and loss expenses, including our inability to realize the remediation recoveries or future commutations included in our reserves; (10) adverse developments in our portfolio of insured public finance credits; (11) market risks impacting assets in our investment portfolio or the value of our assets posted as collateral in respect of investment agreements and interest rate swap and currency swap transactions; (12) risks relating to determination of amount of impairments taken on investments; (13) credit and liquidity risks due to unscheduled and unanticipated withdrawals on investment agreements; (14) market spreads and pricing on insured collateralized loan obligations ("CLOs") and other derivative products insured or issued by Ambac or its subsidiaries; (15) Ambac's financial position and the Segregated Account Rehabilitation Proceedings may prompt departures of key employees and may impact our ability to attract qualified executives and employees; (16) the risk of litigation and regulatory inquiries or investigations, and the risk of adverse outcomes in connection therewith, which could have a material adverse effect on our business, operations, financial position, profitability or cash flows; (17) credit risk throughout our business, including but not limited to credit risk related to residential mortgage-backed securities, student loan and other asset securitizations, CLOs, public finance obligations and exposures to reinsurers; (18) default by one or more of Ambac Assurance's portfolio investments, insured issuers, counterparties or reinsurers; (19) the risk that our risk management policies and practices do not anticipate certain risks and/or the magnitude of potential for loss as a result of unforeseen risks; (20) factors that may influence the amount of installment premiums paid to Ambac, including the Segregated Account Rehabilitation Proceedings; (21) changes in prevailing interest rates; (22) the risk of volatility in income and earnings, including volatility due to the application of fair value accounting, required under the relevant derivative accounting guidance; (23) changes in accounting principles or practices that may impact Ambac's reported financial results; (24) legislative and regulatory developments; (25) operational risks, including with respect to internal processes, risk models, systems and employees; (26) changes in tax laws, tax disputes and other tax-related risks; and (27) other risks and uncertainties that have not been identified at this time. OVERVIEW Ambac Financial Group, Inc. ("Ambac" or the "Company"), headquartered in New York City, is a holding company incorporated in the state of Delaware. Ambac was incorporated on April 29, 1991. On November 8, 2010, Ambac filed a voluntary petition for relief under Chapter 11 of the United States Bankruptcy Code ("Bankruptcy Code") in the United States Bankruptcy Court for the Southern District of New York ("Bankruptcy Court"). Ambac has continued to operate in the ordinary course of business as "debtor-in-possession" under the jurisdiction of the Bankruptcy Court and in accordance with the applicable provisions of the Bankruptcy Code and the orders of the Bankruptcy Court. The Company, as debtor and debtor-in-possession, filed a Fifth Amended Plan of Reorganization on March 12, 2012 (such Fifth Amended Plan of Reorganization, as it may be amended, the "Reorganization Plan"). The Bankruptcy Court entered an order confirming the Reorganization Plan on March 14, 2012. Under the Reorganization Plan, Ambac's debt holders and other creditors will receive all of the equity in the reorganized company. Additionally, the Reorganization Plan sets forth the revised capital structure of a newly reorganized Ambac and provides for corporate governance subsequent to emergence from bankruptcy. The Reorganization Plan is not yet effective. Effectiveness depends on the satisfaction of numerous conditions, including the final settlement of certain disputes. 63-------------------------------------------------------------------------------- Table of Contents Ambac Assurance is Ambac's principal operating subsidiary. Ambac Assurance is a financial guarantee insurer which provided financial guarantees and financial services to clients in both the public and private sectors around the world. In March 2010, Ambac Assurance established a segregated account pursuant to Wisc. Stat. §611.24(2) (the "Segregated Account") to segregate certain segments of Ambac Assurance's liabilities. The Office of the Commissioner of Insurance for the State of Wisconsin ("OCI" (which term shall be understood to refer to such office as regulator of Ambac Assurance and to the Commissioner of Insurance for the State of Wisconsin as rehabilitator of the Segregated Account (the "Rehabilitator"), as the context requires)) commenced rehabilitation proceedings with respect to the Segregated Account (the "Segregated Account Rehabilitation Proceedings") in order to permit the OCI to facilitate an orderly run-off and/or settlement of the liabilities allocated to the Segregated Account pursuant to the provisions of the Wisconsin Insurers Rehabilitation and Liquidation Act. The Rehabilitator is Theodore Nickel, the Commissioner of Insurance of the State of Wisconsin. Ambac Assurance is not, itself, in rehabilitation proceedings. Through its financial services subsidiaries, Ambac provided financial and investment products, including investment agreements, funding conduits and interest rate and currency swaps, principally to the clients of its financial guarantee business. Ambac Assurance insured all of the obligations of its financial services subsidiaries. These businesses are in active runoff, which is being effectuated by means of transaction terminations, settlements, assignments and scheduled amortization of contracts. Financial information concerning our business segments is set forth in the unaudited consolidated financial statements and the notes, "Management's Discussion and Analysis of Financial Condition and Results of Operations," and "Quantitative and Qualitative Disclosures About Market Risk," which are in Part I of this Quarterly Report on Form 10-Q. Our Internet address is www.ambac.com. We make available free of charge, on or through the investor relations section of our web site, annual reports on Form 10-K, quarterly reports on Form 10-Q and current reports on Form 8-K and amendments to those reports filed or furnished pursuant to Section 13(a) or 15(d) of the Securities Exchange Act of 1934, as well as proxy statements, as soon as reasonably practicable after we electronically file such material with, or furnish it to, the U.S. Securities and Exchange Commission. Our Investor Relations Department can be contacted at Ambac Financial Group, Inc., One State Street Plaza, New York, New York 10004, Attn: Investor Relations, telephone: 212-208-3222. CRITICAL ACCOUNTING POLICIES AND ESTIMATES The discussion and analysis of our financial condition and results of operations is based upon our Unaudited Consolidated Financial Statements, which have been prepared in accordance with U.S. generally accepted accounting principles ("GAAP"). The preparation of these financial statements requires us to make estimates and judgments that affect the reported amount of assets and liabilities, revenues and expenses and related disclosure of contingent assets and liabilities at the date of our financial statements. Actual results may differ from these estimates under different assumptions or conditions. Critical accounting policies are considered critical because they place significant importance on management to make difficult, complex or subjective estimates regarding matters that are inherently uncertain. Financial results could be materially different if alternative methodologies were used or if management modified its assumptions or estimates. Management has identified (i) the accounting for loss and loss expenses of non-derivative financial guarantees, (ii) the valuation of financial instruments, including the determination of whether an investment impairment is other-than-temporary and the (iii) valuation allowance on deferred tax assets, as critical accounting policies. This discussion should be read in conjunction with the Unaudited Consolidated Financial Statements and notes thereon included elsewhere in this report. We have discussed with the Audit Committee management's assessment of such critical accounting policies, the reasons why they are considered critical, and how current and anticipated future events impact those determinations. The Company's critical accounting policies and estimates are as follows: Losses and Loss Expenses of Non-derivative Financial Guarantees: The loss and loss expense reserves for financial guarantee insurance discussed herein relates only to Ambac's non-derivative insurance business for insurance policies issued to beneficiaries, including variable interest entities ("VIEs"), for which we do not consolidate the VIE. Losses and loss expenses are based upon estimates of the ultimate aggregate losses inherent in the non-derivative financial guarantee portfolio as of the reporting date. The evaluation process for determining the level of reserves is subject to certain estimates and judgments. ASC Topic 944, Financial Services-Insurance provides guidance for financial guarantee insurance contracts issued by insurance enterprises, including the recognition and measurement of claim liabilities (i.e. loss reserves). Under ASC Topic 944, a loss reserve is recorded on the balance sheet for the excess of (a) the present value of expected losses, over (b) the unearned premium reserve ("UPR") for that contract. Expected losses represent projected net outflows and are defined as the expected future claims to be paid under an insurance contract plus the impact of potential settlement outcomes upon future installment premiums, less potential recoveries. To the extent (a) is less than (b), no loss reserve is recorded. Changes to the loss reserve estimate in subsequent periods are recorded as a loss and loss expense on the Consolidated Statements of Total Comprehensive Income. For those policies where the potential recovery is less than the expected future claims, the resulting net outflow is recorded as a "Loss and loss expense reserve" liability. For those policies where losses have been paid, but not yet recovered, the potential recovery may be greater than the expected future claims and the resulting net inflow is recorded as a "Subrogation recoverable" asset. 64-------------------------------------------------------------------------------- Table of Contents Ambac's loss reserves are based on management's on-going review of the non-derivative financial guarantee insurance portfolio. Active surveillance of the insured portfolio enables Ambac's surveillance group to track credit migration of insured obligations from period to period and update internal credit ratings for each transaction. Non-adversely classified credits are assigned a Class I or Survey List ("SL") risk classification, while adversely classified credits are assigned a risk classification of Class IA through Class V. The criteria for an exposure to be assigned an adversely classified credit rating includes the deterioration of an issuer's financial condition, underperformance of the underlying collateral (for collateral dependent transactions such as mortgage-backed securitizations), poor performance by the servicer of the underlying collateral and other adverse economic events or trends. The servicer of the underlying collateral of an insured securitization transaction is a consideration in assessing credit quality because the servicer's performance can directly impact the performance of the related issue. For example, a servicer of a mortgage-backed securitization that does not remain current in its collection and loss mitigation efforts could cause an increase in delinquencies and potential defaults of the underlying obligations. Similarly, loss severities increase when a servicer does not effectively handle loss mitigation activities such as (i) the advancing of delinquent principal and interest and default-related expenses which are deemed to be recoverable by the servicer, (ii) pursuit of loan charge-offs which maximize cash flows from pool's securitized assets, and (iii) other asset disposition strategies and timelines. The population of credits evaluated in Ambac's loss reserve process are (i) all adversely classified credits and (ii) non-adversely classified credits which had an internal Ambac credit rating downgrade since the transaction's inception. One of two approaches is then utilized to estimate expected losses to ultimately determine if a loss reserve should be established. The first approach is a statistical expected loss approach, which considers the likelihood of all possible outcomes. The "base case" statistical expected loss is the product of: (i) the net par outstanding on the credit; (ii) internally developed historical default information (taking into consideration internal ratings and average life of an obligation); (iii) internally developed loss severities; and (iv) a discount factor. The loss severities and default information are based on rating agency information, are specific to each bond type and were established and approved by Ambac's senior management. For certain credit exposures, Ambac's additional monitoring loss remediation efforts and probabilities of potential settlement outcomes may provide information relevant to adjust this estimate of "base case" statistical expected losses. As such, the loss severities used in estimating the "base case" statistical expected losses may be adjusted based on the professional judgment of the surveillance analyst monitoring the credit with the approval of senior management. Analysts may accept the "base case" statistical expected loss as the best estimate of expected loss or assign multiple probability weighted severities to determine an adjusted statistical expected loss that better reflects a given transaction's potential severity, as well as the potential for additional remediation activities. The second approach entails the use of more precise estimates of future claim payments, net of potential recoveries, arising as a result of the non-performance of the insured exposure and expected to be paid to the holder of the insured financial obligation. Ambac's surveillance group will consider the likelihood of all possible outcomes and develop appropriate cash flow scenarios. This approach can include the utilization of internal or external models to project future claim payment estimates. We have utilized external models for residential mortgage-backed and student loan exposures. In general, these models use historical performance of the collateral pools in order to then assume or derive future performance characteristics, such as default and voluntary prepayment rates, which in turn determine projected future claim payments. In this approach, a probability-weighted expected loss estimate is developed based on assigning probabilities to multiple claim payment scenarios and applying an appropriate discount factor. Additionally, we assign a probability to the issuer's ability to refinance an insured issue and/or Ambac's ability to execute a potential settlement (i.e. commutation) of the insurance policy, inclusive of the impact on future installment premiums. The methodology used to estimate the most substantial amount of the potential recovery component of expected losses is further described in the "RMBS Representation and Warranty Subrogation Recovery" section below. The discount factor applied to both of the above described approaches is based on a risk-free discount rate corresponding to the remaining expected weighted-average life of the exposure and the exposure currency. The discount factor is updated for the current risk-free rate each reporting period. As further described in Note 1 of the Unaudited Consolidated Financial Statements included Part I, Item 1 of this Form 10-Q, no decision has been announced with respect to effectuating or amending the Segregated Account Rehabilitation Plan. Possible modifications to the Plan, with respect to the form and timing of satisfying the remaining balance of unpaid claims, could have a material effect on the measurement of loss and loss expense reserves. Ambac establishes loss expense reserves based on our estimate of expected net cash outflows for loss expenses such as legal and consulting costs. As the probability of default for an individual credit increases and/or the severity of loss given a default increases, our loss reserve for that insured obligation will also increase. Political, economic, credit or other unforeseen events could have an adverse impact on default probabilities and loss severities. The performance and loss reserves for many transactions (such as many public finance exposures) are derived from the issuer's obligation to pay. The performance and loss reserves of other transactions such as 65-------------------------------------------------------------------------------- Table of Contents most structured finance exposures including RMBS have no direct issuer support and therefore are derived from the default activity and loss given default of collateral supporting the transactions. Many transactions have a combination of issuer/entity and collateral support. Loss reserves reflect our assessment of the transaction's overall structure, support and expected performance. Loss reserve volatility will be a direct result of the credit performance of our insured portfolio, including the number, size, bond types and quality of credits included in our loss reserves as well as our ability to effectively utilize remediation strategies to improve performance or reduce our exposure via commutations. The number and severity of credits included in our loss reserves depend to a large extent on transaction specific attributes, but will generally increase during periods of economic stress and decline during periods of economic stability. Reinsurance recoveries do not have a significant effect on loss reserve volatility because Ambac has little exposure ceded to reinsurers and has received collateral from the majority of its reinsurers. Ambac has exposure to various bond types issued in the debt capital markets. Our experience has shown that, for the majority of bond types, we have not experienced significant claims. We have observed that, with respect to certain bond types, it is reasonably possible that a material change in actual loss severities and defaults could occur over time. In the future, our experience may differ with respect to the types of guaranteed bonds affected or the magnitude of the effect. The bond types that have experienced the most significant claims and loss payments are residential mortgage-backed securities ("RMBS"), student loan securities, healthcare obligations, and collateralized debt obligations ("CDOs"). These four bond types represent 93% of our ever-to-date insurance claims recorded with RMBS comprising 88% of our ever-to-date insurance claims recorded. The table below indicates the number of credits, gross par outstanding and gross loss reserves (including loss adjustment expenses) related to policies in Ambac's loss reserves at September 30, 2012: Number of Gross par Gross Loss ($ in millions) credits outstanding(1) Reserves(1)(2)(3) RMBS 169 $ 14,669 $ 3,901 Student Loans 44 4,541 1,109 All other credits 99 5,675 1,370 Loss adjustment expenses n.a. n.a. 140 Totals 312 $ 24,885 $ 6,520 (1) Ceded Par Outstanding and ceded loss reserves are $1,062 and $166, respectively. Ceded loss reserves are included in Reinsurance Recoverable on paid and unpaid losses. (2) Loss reserves of $6,520 are included in the balance sheet in the following line items: Loss and loss expense reserve-$7,034 and Subrogation recoverable-$514. (3) Included in gross loss reserves are unpaid claims of $3,387 million relating to policies allocated to the Segregated Account. RMBS: Ambac insures RMBS transactions that contain first-lien mortgages. Ambac classifies its insured first-lien RMBS exposure principally into two broad credit risk classes: mid-prime (including Alt-A, interest only, and negative amortization) and sub-prime. Mid-prime loans were typically made to borrowers who had stronger credit profiles relative to sub-prime loans, but weaker than prime loans. Compared with mid-prime loans, sub-prime loans typically had higher loan-to-value ratios, reflecting the greater difficulty that sub-prime borrowers have in making down payments and the propensity of these borrowers to extract equity during refinancing. The mid-prime category includes: • Loans with specific payment features that afforded borrowers the option to have lower payments in the early years with potential resets after several years. For example, so-called interest only loans have monthly payments comprised of interest but no principal. So called negative amortization loans permit borrowers to defer interest and principal in the early years and then make higher payments in the period after the reset. Both types may also have lower interest rates in the early years. Future increases in monthly payments, commonly called payment shock, raise the probability of delinquencies and defaults given the decline in house prices that has caused many borrowers' loan amounts to exceed their homes market value. • Loans backed by borrowers who typically did not meet standard agency guidelines for documentation requirement, property type or loan-to-value ratio. These are typically higher-balance loans made to individuals who might have past credit problems that were not severe enough to warrant "sub-prime" classification, or borrowers who chose not to obtain a prime mortgage due to documentation requirements. Ambac has also insured RMBS transactions that contain predominantly second-lien mortgage loans such as closed-end seconds and home equity lines of credit. A second-lien mortgage loan is a type of loan in which the borrower uses the equity in their home as collateral and the second-lien loan is subordinate to the first-lien loan outstanding on the home. The borrower is obligated to make monthly payments on both their first and second-lien loans. If the borrower defaults on the payments due under these loans and the property is subsequently liquidated, the liquidation proceeds are first utilized to pay off the first-lien loan (as well as costs due the servicer) and any remaining funds are applied to pay off the second-lien loan. As a result of this subordinate position to the first-lien loan, second-lien loans carry a significantly higher severity in the event of a loss, typically at or above 100% in the current housing market. 66 -------------------------------------------------------------------------------- Table of Contents RMBS transaction-specific behavior is analyzed on a risk-priority basis. We employ a screening tool to assess the sufficiency of credit enhancement remaining in a transaction, as well as other adverse credit data that may identify deterioration. Transactions which are experiencing escalating delinquencies and increasing loss severities and/or which are experiencing declining levels of subordination or overcollateralization relative to collateral losses are identified as underperforming. For underperforming transactions, historical collateral performance is examined and future collateral performance and cash flows are projected and evaluated. These underperforming transactions are then included in an adversely classified credit list and assigned a credit classification consistent with the degree of underperformance. Methodology for Projecting Expected Losses in our RMBS Portfolio For the first three quarters of 2011, we determined expected losses by using (i) an internal roll-rate model to project losses for our second-lien transactions; (ii) a licensed third-party multi-scenario stochastic (Monte Carlo) cash flow model ("stochastic model") to project losses for our first-lien transactions; and (iii) a licensed statistical regression model ("regression model") to develop estimates of projected losses for both our second and first-lien transactions. As of the quarter ending December 31, 2011, we discontinued the use of both the stochastic model and the internal roll-rate model and utilized the regression model to develop estimates of projected losses. Our reserving methodology in the first three quarters of 2011 reflected a blending of the results of the two approaches used for each transaction, with 50% probability assigned to the regression model and 50% assigned to either our internal roll-rate model in the case of second-lien transactions or to the stochastic model in the case of our first-lien transactions. In the months leading up to the transition to one model, we worked extensively with the regression model vendor to ensure the model was adequately projecting the performance of our transactions, and to provide additional information and data to improve the precision of the model's projections. Prior to solely utilizing the results of the regression model, our internal RMBS credit and quantitative professionals evaluated and analyzed the results of the regression model versus loss estimates generated utilizing our internal roll-rate model for second-lien transactions and our stochastic model for first-lien transactions. This cross disciplined team compared the specific drivers and methodology of the regression model with our existing approaches, and analyzed deal performance and model outputs across the portfolio. For example, the team considered the general reasonableness of the models' projected defaults of borrowers not currently in seriously delinquent or worse payment status and also conducted selective collateral analyses. The team also assessed the models' cumulative loss estimates and compared such estimates by asset type and vintage with rating agency and other published loss projections. Based upon this analysis, we believe the exclusive use of the regression model to project RMBS losses was reasonable. Although RMBS loss projections can be widely disparate and there can be no certainty with regard to projecting such losses, we believe our current reserving approach, including the regression model itself and the assumptions utilized, is sound and reasonable. The regression model projects losses utilizing: (i) the RMBS transaction's collateral, characteristics and status, (ii) the RMBS transaction's payment waterfall structure, including the application of interest and principal payments and recoveries, (iii) projected home price appreciation ("HPA"), and (iv) projected interest rates. We provide the regression model vendor with both HPA projections from a recognized source and interest rate projections we develop from market sources. We generally utilize waterfall structures from a market accepted vendor of securitization deal structures. In some cases, we may utilize an internally developed waterfall structure when our legal and commercial analysis of the transaction's payment structure differs from the vendor's waterfall structure. We compare monthly claims submitted against the trustees' reports, third-party provided waterfall models and our understanding of the transactions' structures to identify and resolve discrepancies. We also review the vendor's published waterfall revisions to further identify material discrepancies. Resolving discrepancies is challenging and may take place over an extended period of time. Moreover, transaction documents are subject to interpretation, and our interpretation or that of the vendor and as reflected in our loss reserves may prove to be incorrect and/or not executed by the trustees directing cash flows in the future. The regression model is subject to ongoing refinements resulting from industry research and performance that may better inform model assumptions, enhance model capabilities and other factors. For example, an enhancement was made to the model in the first quarter of 2012 to better reflect the potential impact of borrowers' payment histories on default rates. In our experience, market performance and model characteristics change and are updated through time and a regular review of models and the overall approach to loss estimation is beneficial. Our ability to drive change in the models we license is limited and subject to the expertise and views of the independent developer/vendor. On the other hand, our ability to estimate losses without such models is difficult and challenging for a large portfolio across multiple RMBS exposure types. Summarized below is our approach to projecting claims and ultimate losses in our RMBS portfolio. 67 -------------------------------------------------------------------------------- Table of Contents Second-Lien The regression model estimates mortgage loan collateral performance, the effect of such collateral cash flows within the transaction waterfall and the liability structure we insure. Collateral performance is frequently modeled at the deal level given the paucity of mortgage loan level data for second-lien transactions. Without specific loan-level information, the deal-level approach processes a loan pool as if it were a single loan, selecting certain aggregated deal-level characteristics to then perform a statistical analysis using a multinomial logistic regression model. We use three home price appreciation ("HPA") projection scenarios to develop a base case as well as stress and upside cases. The highest probability is assigned to the base case, with significantly lower probabilities to the stress and upside cases. This deal-level approach of the regression model takes a relatively complicated monthly cash flow and simplifies it into two parts: a borrower-behavior-dependent stage and a servicer-behavior-dependent stage. The borrower stage is designed to forecast the probability of a loan's present delinquency status transitioning to any of eight future statuses. Through the borrower-behavior-dependent stage, at any given monthly period, a loan can take on one of the following eight statuses: current, 30, 60, 90, 120, 150, 180+ days delinquent, or prepaid. The deal-level approach calculates defaults using a roll-rate that evaluates the possible future state of a set of loans based on its current status and three variables: average FICO (credit score), average current consolidated loan to value ratio ("CLTV"), and an overall quality indicator. The servicer-behavior-dependent stage of the regression model governs a loan's life cycle after it reaches 180 or more days delinquent. This stage evaluates the servicer's propensity to foreclose or pursue a short sale, the speed of the foreclosure process, and the speed of the post-foreclosure distressed property liquidation. The transition probabilities between advanced delinquency and foreclosure, foreclosure and REO, and finally REO and ultimate liquidation are assumed by the model to depend upon how long a loan has already been in a particular status, as well as on the servicer and on state-specific liquidation timeline factors. First-Lien For most first-lien transactions, the regression model utilizes home mortgage loan level data from recognized market sources to calculate each loan's probability of default and prepayment based on the characteristics of the loan. The loan-level approach of the regression model uses the results of a regression analysis to project prepayment and default vectors on a monthly basis based on its embedded risks. For first-lien transactions that do not have loan-level data available, we use the deal-level approach of the regression model that is described in the Second-Lien section above. There are three transitional stages with the loan-level approach of the regression model: current, prepayment or default. The model then looks beyond the stages to assess a set of loans based on a number of individual characteristics that are distinct to that set of loans. These individual characteristics are property type, occupancy status, loan purpose, documentation type, cumulative loan-to-value ratio, originator quality rating, servicer quality rating, FICO score, original loan balance, interest rate margin, and regional unemployment rate. The model then estimates the rate at which a loan will prepay or default reducing the balance of each loan monthly during the projection period based on the borrower's given probability of defaulting or prepaying for that month. Servicer behavior is a unique variable in the loan-level approach of the regression model and is used to calculate the impact of servicer performance on expected prepayments and defaults. Consistent with the second-lien modeling, we consider three HPA scenarios in the regression model to develop a base case as well as stress and upside cases. The highest probability is assigned to the base case, with significantly lower probabilities to the stress and upside cases. Additional RMBS factors for first and second-lien transactions Additional factors that may impact ultimate RMBS losses include, but may not be limited to, mortgage insurance, government programs and servicer intervention. These factors, and the impact on our loss reserve estimate, are further discussed below: Mortgage insurance: Six of our mortgage-backed transactions have pool-level mortgage insurance remaining. Pool mortgage insurance is a master policy issued to the mortgage securitization trust, which indemnifies the trust either on a first loss or mezzanine basis in the event that covered mortgage loans in the trust default. The mortgage insurance master policy specifies the particular characteristics and conditions of each individual loan within the mortgage trust that is subject to coverage. The policy also includes various conditions including exclusions, conditions for notification of loans in default and claims settlement. We have noted with regard to these transactions that payment by mortgage insurers of claims presented by the mortgage trusts has been inconsistent; resulting in higher claims presented under Ambac Assurance's financial guarantee policies. As a result, the impact of mortgage insurance on our loss reserve estimate is negligible. Government programs: In May of 2009, the Federal Government initiated the Home Affordable Modification Plan (HAMP) which allows servicers to modify loans. After determining a borrower's eligibility, a servicer can take a series of steps to reduce the monthly mortgage payment. HAMP is applicable to the Ambac-wrapped transactions serviced by the servicers that have signed servicer participation agreements to modify loans under HAMP. Based on the activity HAMP offers and indications from government published sources that suggest this program is unlikely to have a material impact on Ambac-wrapped transactions, beginning in the third quarter of 2011 we no longer assumed any impact for HAMP on our first-lien portfolio. At December 31, 2010, Ambac's first-lien stochastic model assumed that 0.5% of HAMP-eligible loans will be modified monthly for 24 months for Ambac portfolios serviced by HAMP participating servicers. During the first half of 2011 we reduced the impact of HAMP modifications to reflect the reported performance of HAMP. We have not given credit to any other government programs, most of which are targeted to home mortgages that are bank owned, and not in RMBS securitizations. 68 -------------------------------------------------------------------------------- Table of Contents Servicer Intervention: With the exception of the internal second-lien roll-rate model used until the fourth quarter of 2011, we are able to include in our modeling the steps which Ambac is taking to address shortcomings in servicing performance including transfers of servicing where the legal right exists to do so. Ambac has initiated, with the cooperation of the Rehabilitator of the Segregated Account, programs with special servicers that we believe will mitigate losses on such transactions through intervention strategies such as loan modifications, improved liquidation timelines, and short sales. Ambac believes these are the principal factors that will result in reduced losses over time. Given the uncertainty in initiating additional programs of this nature, we are projecting that only exposures that have already transferred servicing or entered into special servicing agreements will benefit from the effects of servicer intervention strategies. RMBS Representation and Warranty Subrogation Recoveries: In an effort to better understand the unprecedented levels of delinquencies, Ambac or its counsel engaged consultants with significant mortgage underwriting experience to review the underwriting documentation for mortgage loans underlying certain insured RMBS transactions. Transactions which exhibited exceptionally poor performance were chosen for further examination of the underwriting documentation supporting the underlying loans. Factors which Ambac believes to be indicative of poor performance include (i) high levels of early payment defaults, (ii) significant number of loan liquidations or charge-offs and resulting high levels of losses, and (iii) rapid elimination of credit protections inherent in the transactions' structures. Item (ii), "loan liquidations" refers to loans for which the servicer has liquidated the related collateral and the securitization has realized losses on the loan; "charge-offs" refers to loans which have been written off as uncollectible by the servicer, thereby generating no recoveries to the securitization, and may also refer to the unrecovered balance of liquidated loans. In either case, the servicer has taken actions to recover against the collateral, and the securitization has incurred losses to the extent such actions did not result in full repayment of the borrower's obligations. Generally, the sponsor of the transaction provided representations and warranties with respect to the securitized loans, including representations with respect to the loan characteristics, the absence of fraud or other misconduct in the origination process, and attesting to the compliance of loans with the prevailing underwriting policies. In such cases, the sponsor of the transaction is contractually obligated to repurchase, cure or substitute collateral for any loan that breaches the representations and warranties. Refer to Note 6 of the Unaudited Consolidated Financial Statements included Part I, Item 1 of this Form 10-Q for more information regarding representation and warranty subrogation recoveries. The table below distinguishes between RMBS credits for which we have not established a representation and warranty subrogation recovery and those for which we have, providing in both cases the number of policies, gross par outstanding, gross loss reserves before subrogation recoveries, subrogation recoveries, and gross loss reserves net of subrogation for all RMBS exposures for which Ambac established reserves at September 30, 2012: Gross loss reserve Gross loss Gross before reserve net of Number of par subrogation Subrogation subrogation ($ in millions) policies outstanding recoveries recoveries recoveries Second-lien 22 $ 2,206 $ 553 $ - $ 553 First-lien-Mid-prime 58 3,800 1,771 - 1,771 First-lien-Sub-prime 34 1,400 206 - 206 Other 11 338 214 - 214 Total Credits Without Subrogation 125 7,744 2,744 - 2,744 Second-lien 23 3,366 1,556 (1,546 ) 10 First-lien Mid-prime 16 1,476 1,181 (655 ) 526 First-lien Sub-prime 5 2,083 1,109 (488 ) 621 Total Credits With Subrogation 44 6,925 3,846 (2,689 ) 1,157 Total 169 $ 14,669 $ 6,590 $ (2,689 ) $ 3,901 STUDENT LOANS: Our student loan portfolio consists of credits collateralized by (i) federally guaranteed loans under the Federal Family Education Loan Program ("FFELP") and (ii) private student loans. Whereas FFELP loans are guaranteed for a minimum of 97% of defaulted principal and interest, private loans have no government guarantee and therefore are subject to credit risk as with other types of unguaranteed credits. Private student loans are outside the purview of recent government programs designed to assist borrowers. Recent default data has shown a continued deterioration in the performance of private student loans underlying our transactions. Due to the failure of the auction rate and variable rate markets in 2008, the interest rates on the Ambac insured securities increased 69 -------------------------------------------------------------------------------- Table of Contents significantly to punitive levels pursuant to the terms of the transactions. Such increases have caused the collateralization ratio in these transactions to deteriorate due to negative excess spread and/or the use of principal receipts to pay current interest. Further rating agency downgrades of outstanding auction rate notes has resulted in a step-ups of the interest rate payable on such securities which further accelerate the erosion of the trust estate. Methodology for Projecting Expected Losses in our Student Loan Portfolio The calculation of loss reserves for our student loan portfolio involves evaluating numerous factors that can impact ultimate losses. The factor which contributes to the greatest degree of uncertainty in ascertaining appropriate loss reserves is the long time horizon associated with the final legal maturity date of the bonds. Most of the student loan bonds which we insure were issued with original terms of 20 to 40 years until final maturity. Since our policy covers timely interest and ultimate payment of principal, our loss projections must make assumptions for many factors covering a long time horizon. Key assumptions that will impact ultimate losses include but are not limited to the following: collateral performance which is highly correlated to the economic environment, interest rates, operating risks associated with the issuer, servicers, administrators, issuers willingness and ability to refinance, investor appetitive for tendering Auction Rate Securities at a discount and, as applicable, Ambac's ability and willingness to commute policies. In evaluating our Student Loan portfolio, losses were projected using either a cash flow or statistical expected loss approach. As noted above, the statistical methodology uses probability of default and loss given default (LGD) to derive a weighted average loss expectation. The scenarios used under the statistical expected loss approach evaluate each transaction under base case and stress case scenarios. The main drivers in assigning appropriate probabilities to LGDs for each policy includes an analysis of the collateral mix; debt type and interest rates; parity level; enhancement levels and remediation opportunities. We believe the statistical expected loss approach is a more efficient methodology for certain deals in our student loan portfolio, such as (i) deals where collateral loan level data is unavailable (and thus the cash flow model, as more fully discussed below could not be used), and (ii) deals where we do not expect to experience meaningful losses. We use a third party deterministic cash flow model to develop loss projections for a portion of our portfolio. The model allows us to capture each transaction's particular structure (i.e. the waterfall structure, triggers, redemption priority). For collateral performance, the model uses loan level detail that allows us to make specific default and recovery assumptions for each type of loan. We contract with a separate third party to run the model at our direction. We provide the third party with deal level assumptions such as default, recovery and interest rate assumptions. We develop multiple cash flow scenarios and assign probabilities to each cash flow scenario based on each transaction's unique situation. Probabilities assigned are based on available data related to the credit, any contact with the issuer, and any economic or market information that may impact the outcomes of the various scenarios being evaluated. Our base case usually projects the deal out to maturity using expected loss assumptions and interest rates adhering to the projected forward interest rate curve at the reporting date. We also develop stress cases that incorporate various stresses to the transaction, including but not limited to defaults, recoveries and interest rates. In estimating loss reserves, we also incorporate scenarios which represent remediation strategies. Remediation scenarios may include the following; (i) a potential refinancing of the transactions by the issuer; (ii) the issuer's ability to redeem outstanding auction rate securities at a discount, thereby increasing the structure's ability to absorb future losses; and (iii) our ability to terminate the policy in whole or in part (e.g. commutation). The remediation scenarios and the related probabilities of occurrence vary by policy depending on on-going discussions and negotiations that are underway with issuers and/or investors. In addition to commutation negotiations that are underway with various counterparties in various forms, our reserve estimates may also include scenarios which incorporate our ability to commute additional exposure with other counterparties. REASONABLY POSSIBLE ADDITIONAL LOSSES: RMBS: It is possible that our loss estimate assumptions for the RMBS insurance policies discussed above could be materially under-estimated as a result of various uncertainties including, but not limited to, continued deterioration in housing prices, poor servicing, rising interest rates on variable rate transactions, the effects of a weakened economy marked by growing unemployment and wage pressures, inability to execute commutation transactions with issuers and/or investors and/or continued illiquidity of the mortgage market. Additionally, our actual subrogation recoveries could be lower than our current estimates if the sponsors of these transactions: (i) fail to honor their obligations to repurchase the mortgage loans, (ii) successfully dispute our breach findings, or (iii) no longer have the financial means to fully satisfy their obligations under the transaction documents. We have attempted to identify the reasonably possible additional losses using more stressful assumptions. Different methodologies, assumptions and models could produce different base and reasonably possible additional losses and actual results may differ materially from any of these various modeled results. In the case of both first and second-lien exposures, the regression model's reasonably possible stress case assumes a significantly harsher HPA projection, which in turn drives higher defaults and severities. Using this approach, the reasonably possible increase in loss reserves for RMBS credits for which we have an estimate of expected loss at September 30, 2012 could be approximately $881 million. The reasonably possible scenario considers the highest stress scenario that was utilized in the development of our probability-weighted expected loss at September 30, 2012 and assumes an inability to execute commutation transactions. 70 -------------------------------------------------------------------------------- Table of Contents Student Loans: It is possible that our loss estimate assumptions for student loan credits could be materially under-estimated as a result of various uncertainties including but not limited to, the interest rate environment, an increase in default rates and loss severities on the collateral due to economic factors, inadequate servicing, inability to execute commutation transactions with issuers and/or investors, as well as a failure of issuers to refinance insured bonds which have a failed debt structure, such as auction rate securities and variable rate debt obligations. For student loan credits for which we have an estimate of expected loss at September 30, 2012, the reasonably possible increase in loss reserves from loss reserves at September 30, 2012 could be approximately $1,199 million. The reasonably possible scenario considers the highest stress scenario that was utilized in the development of our probability-weighted expected loss at September 30, 2012 and assumes an inability to execute commutation transactions. Other Credits: It is possible our loss reserves on other types of credits may be materially under-estimated because most credits have possible outcomes more adverse than the recorded probability weighted loss reserve. Although we do not believe it is reasonably possible to have worst case outcomes in all cases, it is reasonably possible we could have worst case outcomes in some or even many cases. Similarly, it is also reasonably possible we will achieve better outcomes than our recorded probability weighted loss reserve in some cases. For all credits with loss reserves at September 30, 2012, the sum of all the highest stress case loss scenarios is $1,334 million greater than the current loss reserve. The reasonably possible scenario assumes an inability to execute commutation transactions. Ambac's management believes that the reserves for losses and loss expenses and unearned premium reserves are adequate to cover the ultimate net cost of claims, but reserves are based on estimates and there can be no assurance that the ultimate liability for losses will not exceed such estimates. Valuation of Financial Instruments: Ambac's financial instruments that are reported on the Consolidated Balance Sheets at fair value and subject to valuation estimates include investments in fixed income securities, VIE assets and liabilities and derivatives comprising credit default, interest rate and currency swap transactions. Surplus notes issued by Ambac Assurance or the Segregated Account of Ambac Assurance are recorded at fair value at the date of issuance and subsequently reported at amortized cost within Long-term debt on the Consolidated Balance Sheets. Determination of fair value for newly issued surplus notes is a highly subjective process which relies upon the use of significant unobservable inputs and management judgment consistent with a Level 3 valuation. The fair market values of financial instruments held are determined by using independent market quotes when available and valuation models when market quotes are not available. ASC Topic 820, Fair Value Measurements and Disclosures requires the categorization of these assets and liabilities according to a fair value valuation hierarchy. Approximately 85% of our assets and approximately 76% of our liabilities are carried at fair value and categorized in either Level 2 of the valuation hierarchy (meaning that their fair value was determined by reference to quoted prices for similar instruments in active markets, quoted prices for identical or similar instruments in inactive markets and other observable inputs) or Level 3 (meaning that their fair value was determined by reference to significant inputs that are unobservable in the market and therefore require a greater degree of management judgment). The determination of fair value for financial instruments categorized in Level 2 or 3 involves significant judgment due to the complexity of factors contributing to the valuation. Third-party sources from which we obtain independent market quotes also use assumptions, judgments and estimates in determining financial instrument values and different third parties may use different methodologies or provide different prices for securities. In addition, the use of internal valuation models for certain highly structured instruments, such as credit default swaps, require assumptions about markets in which there has been a negligible amount of trading activity for several years. As a result of these factors, the actual trade value of a financial instrument in the market, or exit value of a financial instrument position owned by Ambac, may be significantly different from its recorded fair value. Refer to Note 7 to the Unaudited Consolidated Financial Statements in Part I, Item 1 of this Form 10-Q for discussion related to the transfers in and/or out of Level 1, 2 and 3 fair value categories. Investment in Fixed Income Securities: Investments in fixed income securities are accounted for in accordance with ASC Topic 320, Investments-Debt and Equity Securities. ASC Topic 320 requires that all debt instruments and certain equity instruments be classified in Ambac's Consolidated Balance Sheets according to their purpose and, depending on that classification, be carried at either cost or fair market value. The fair values of fixed income investments held in the investment portfolios of Ambac and its operating subsidiaries are based primarily on quoted market prices received from dealer quotes or alternative pricing sources with reasonable levels of price transparency. For those fixed income investments where quotes were not available, fair values are based on internal valuation models. Refer to Note 7 to the Unaudited Consolidated Financial Statements in Part I, Item 1 of this Form 10-Q for further discussion of the valuation methods, 71-------------------------------------------------------------------------------- Table of Contents inputs and assumptions for fixed income securities. Ambac performs various review and validation procedures to quoted and modeled prices for fixed income securities, including price variance analyses, missing and static price reviews, overall valuation analyses by senior traders and finance managers and reviews associated with our ongoing impairment analysis. Unusual prices identified through these procedures will be evaluated further against separate broker quotes (if available) or internally modeled prices, and the pricing source values will be challenged as necessary. Price challenges generally result in the use of the pricing source's quote as originally provided or as revised by the source following their internal diligence process. A price challenge may result in a determination that the pricing source cannot provide a reasonable value for a security or cannot adequately support a quote, in which case Ambac would resort to using either other quotes or internal models. Results of price challenges are reviewed and approved by senior traders and finance managers. Valuation results, particularly those derived from valuation models and quotes on certain mortgage and asset-backed securities, could differ materially from amounts that would actually be realized in the market. Ambac's investments in fixed income securities (excluding VIE investments) classified as "available-for-sale" are carried at fair value, with the after-tax difference from amortized cost reflected in stockholders' equity as a component of Accumulated Other Comprehensive Income ("AOCI"). One of the significant estimates related to available-for-sale securities is the evaluation of investments for other-than-temporary impairments. If management assesses that it either (i) has the intent to sell its investment in a debt security or (ii) more likely than not will be required to sell the debt security before the anticipated recovery of its amortized cost basis less any current period credit loss, then an other-than-temporary impairment charge must be recognized in earnings, with the amortized cost of the security being written-down to fair value. If these conditions are not met, but it is determined that a credit loss exists, the impairment is separated into the amount related to the credit loss, which is recognized in earnings, and the amount related to all other factors, which is recognized in other comprehensive income. To determine whether a credit loss has occurred, management considers certain factors, including the length of time and extent to which the fair value of the security has been less than its amortized cost and downgrades of the security's credit rating. If such factors indicate that a potential credit loss exists, management will compare the present value of estimated cash flows from the security to the amortized cost basis to assess whether the entire amortized cost basis will be recovered. If it is determined that all or a portion of the amortized cost basis will not be recovered, a credit impairment charge is recorded in earnings in the amount of the difference between the present value of cash flows and the amortized cost at the balance sheet date, with the amortized cost basis of the impaired security written-down to the present value of cash flows. Ambac estimates expected future cash flows from residential mortgage-backed securities using models and assumptions consistent with those used to project losses in the financial guarantee RMBS portfolio described above under "Critical Accounting Policies-Losses and Loss Expenses of Non-derivative Financial Guarantees." Estimated cash flows are discounted at the effective interest rate implicit in the security at the date of acquisition or, for debt securities that are beneficial interests in securitized financial assets, at a rate equal to the current yield used to accrete the beneficial interest. For floating rate securities, estimated cash flows are projected using the relevant index rate forward curve and the discount rate is adjusted for changes in that curve. For debt securities for which other-than-temporary impairments were recognized in earnings, the difference between the new amortized cost basis and the cash flows expected to be collected are accreted as interest income over the expected remaining life of the security. Ambac's investment portfolio includes certain securities that are guaranteed by Ambac Assurance. As described further in Note 8 to the Unaudited Consolidated Financial Statements in Part I, Item 1 of this Form 10-Q, future cash flows used to measure credit impairment of Ambac-wrapped bonds represents the sum of (i) the bond's intrinsic cash flows and (ii) the estimated fair value of Ambac claim payments. Under the Segregated Account Rehabilitation Plan, which has been confirmed but is not effective and is subject to change, future claim payments made by Ambac Assurance on these securities would be satisfied 25% in cash and 75% in surplus notes. It is uncertain whether the actual form and amount of claim payments will conform to that set forth in the Segregated Account Rehabilitation Plan. The Rehabilitator may seek to effectuate the current Segregated Account Rehabilitation Plan, modify such Plan or modify the injunctions issued by the Rehabilitation Court to allow for the payment of policy claims in such manner and at such times as the Rehabilitator determines to be in the best interest of policyholders. Pursuant to the injunctions issued by the Rehabilitation Court, claims on policies allocated to the Segregated Account were not paid from the commencement of the Segregated Account Rehabilitation Proceedings until the third quarter of 2012. As further described in Note 1 of the Unaudited Consolidated Financial Statements included in Part I, Item 1 of this Form 10-Q, on September 20, 2012, in accordance with published Policy Claim Rules, the Segregated Account commenced paying 25% of each permitted policy claim that arose since the commencement of the Segregated Account Rehabilitation Proceedings. The Segregated Account will continue to make cash payments of 25% of each policy claim submitted and permitted in accordance with such Policy Claim Rules, subject to any further orders of the Rehabilitation Court. As further described in Note 1 of Part 1 - Notes to the Unaudited Consolidated Financial Statements, no decision has been announced with respect to effectuating or amending the Segregated Account Rehabilitation Plan. Possible modifications to the Plan, with respect to the form and timing of satisfying the remaining balance of unpaid claims, could have a material effect on the recognition of other-than-temporary impairment for Ambac-wrapped securities. The Rehabilitator has previously announced that more specific information regarding the status of the Segregated Account Rehabilitation Plan, including possible modifications, will be provided as soon as appropriate. Estimation of the fair value of future claim payments from the Segregated Account is highly subjective. The evaluation of securities for impairments is a quantitative and qualitative process, which is subject to risks and uncertainties and is intended to determine whether declines in the fair value of investments should be recognized in current period earnings. The risks and uncertainties include changes in general economic conditions, the issuer's or guarantor's financial condition and/or future prospects, the effects of changes in interest rates or credit spreads and the expected recovery period. There is also significant judgment 72 -------------------------------------------------------------------------------- Table of Contents in determining whether Ambac intends to sell securities or will continue to have the ability to hold temporarily impaired securities until recovery. Future events could occur that were not reasonably foreseen at the time management rendered its judgment on the Company's intent to retain such securities until recovery. Examples of such events include, but are not limited to, the deterioration in the issuer's or guarantor's creditworthiness, a change in regulatory requirements or a major business combination or major disposition. VIE Assets and Liabilities: The financial assets and liabilities of VIEs consolidated under ASC Topic 810, Consolidationconsist primarily of fixed income securities, loans receivable, derivative instruments and debt instruments and are generally carried at fair value with changes in fair value recognized in Income on variable interest entities of the Consolidated Statements of Total Comprehensive Income. These consolidated VIEs are primarily securitization entities which have liabilities and/or assets guaranteed by Ambac Assurance or Ambac UK. The fair values of VIE debt instruments are determined using the same methodologies used to value Ambac's fixed income securities in its investment portfolio as described above. VIE derivative asset and liability fair values are determined using valuation models. When specific derivative contractual terms are available and may be valued primarily by reference to interest rates, exchange rates and yield curves that are observable and regularly quoted, the derivatives are valued using vendor-developed models. Other derivatives within the VIEs that include significant unobservable valuation inputs are valued using internally developed models. The fair value of VIE assets are obtained from market quotes when available. Typically the asset fair values are not readily available from market quotes and are estimated internally. The consolidated VIEs are securitization entities in which net cash flows from assets and derivatives (after adjusting for financial guarantor cash flows and other expenses) will be paid out to note holders or equity interests. Therefore, when market quotes are not available, our valuation of VIE assets (fixed income securities or loans) are derived from the fair value of debt and derivatives, as described above, adjusted for the fair value of cash flows related to the financial guarantee. The fair value of financial guarantee cash flows include: (i) estimated future premiums discounted at a rate consistent with that implicit in the fair value of the VIE's liabilities and (ii) internal estimates of future loss payments by Ambac Assurance or Ambac UK discounted to consider the guarantor's credit risk. Derivatives: Ambac's operating subsidiaries' exposure to derivative instruments is created primarily through interest rate swaps, US Treasury futures contracts and credit default swaps. These contracts are accounted for at fair value under ASC Topic 815, Derivatives and Hedging. Valuation models are used for the derivatives portfolios, using market data from a variety of third-party data sources. Several of the more significant types of market data that influence fair value include interest rates (taxable and tax-exempt), credit spreads, default probabilities, recovery rates, comparable securities with observable pricing, and the credit rating of the referenced entities. The valuation of certain interest rate and currency swaps as well as all credit derivative contracts also require the use of data inputs and assumptions that are determined by management and are not readily observable in the market, including the amount that Ambac's own credit risk impacts the fair value of derivative liabilities. Refer to Note 7 to the Unaudited Consolidated Financial Statements in Part I, Item 1 of this Form 10-Q for further discussion of the models, model inputs and assumptions used to value derivative instruments. Due to the inherent uncertainties of the assumptions used in the valuation models to determine the fair value of derivative instruments, actual value realized in a market transaction may differ significantly from the estimates reflected in our financial statements. Ambac's credit derivative valuation model, like any financial model, has certain strengths and weaknesses. We believe our model's primary strength is that it maximizes the use of market-driven inputs. Most importantly, the model uses market-based discount rates and fair values of the underlying reference obligations. Ambac employs a three-level hierarchy for obtaining reference obligation fair values used in the model as follows: (i) broker quotes on the reference obligation, (ii) prices and spreads from other transactions in the portfolio with similar asset, structure and credit attributes, and (iii) internal models comparable to those used for invested assets when quotes are unavailable. We believe using this type of approach is preferable to other models, which may emphasize modeled expected losses or which rely more heavily on the use of market indices that may not be reflective of the underlying reference obligation. Another strength is that our model is relatively easy to understand, which increases its transparency. A potential weakness of our valuation model is our reliance on broker quotes obtained from dealers which originated the underlying transactions, who in certain cases may also be the counterparty to our CDS transaction. All of the transactions falling into this category are illiquid and it is usually difficult to obtain alternative quotes. Ambac employs various procedures to corroborate the reasonableness of quotes received; including comparing to other quotes received on similarly structured transactions, observed spreads on structured products with comparable underlying assets and, on a selective basis when possible, values derived through internal estimates of discounted future cash flows. Each quarter, the portfolio of CDS transactions is reviewed to ensure every reference obligation price has been updated. Period to period valuations are compared for each CDS and by underlying bond type. For each CDS, this analysis includes comparisons of key valuation inputs to the prior period and against other CDS within the bond type. No adjustments were made to the broker quotes we received when determining fair value of CDS contracts as of September 30, 2012. 73 -------------------------------------------------------------------------------- Table of Contents Another potential weakness of our valuation model is the lack of new CDS transactions executed by financial guarantors, which makes it difficult to validate the percentage of the reference obligation spread which would be captured as a CDS fee at the valuation date (i.e. the relative change ratio). Changes to the relative change ratio based on internal ratings assigned are another potential weakness as internal ratings could differ from actual ratings provided by independent rating agencies. However, we believe our internal ratings are updated at least as frequently as the external ratings. We believe the approach we have developed to increase the relative change ratio as the underlying reference obligation experiences credit deterioration is consistent with a market-based approach to valuation. Ultimately, our approach exhibits the same weakness as other modeling approaches, as it is unclear if we could execute at these values. Valuation of Deferred Tax Assets: Our provision for taxes is based on our income, statutory tax rates and tax planning opportunities available to us in the jurisdictions in which we operate. Tax laws are complex and subject to different interpretations by the taxpayer and respective governmental taxing authorities. Significant judgment is required in determining our tax expense and in evaluating our tax positions. We review our tax positions quarterly and adjust the balances as new information becomes available. Deferred tax assets arise because of temporary differences between the financial reporting and tax bases of assets and liabilities, as well as from net operating loss and tax credit carry forwards. More specifically, deferred tax assets represent net operating loss carry forwards and current years differences between deductions and losses recorded under U.S. GAAP in a current period which are only deductible for tax purposes in future periods. In accordance with ASC Topic 740, Income Taxes, we evaluate our deferred income taxes quarterly to determine if valuation allowances are required. ASC Topic 740 requires that companies assess whether valuation allowances should be established against their deferred tax assets based on the consideration of all available evidence using a "more likely than not" standard. All available evidence, both positive and negative, needs to be identified and considered in making the determination with significant weight given to evidence that can be objectively verified. The level of deferred tax asset recognition is influenced by management's assessment of future expected taxable income, which depends on the existence of sufficient taxable income of the appropriate character (ordinary vs. capital) within the carry back or carry forward periods available under the tax law. In the event that we determine that we would not be able to realize all or a portion of our deferred tax assets, we would record a valuation allowance against the deferred tax assets that we estimate will not ultimately be recoverable in the period in which that determination is made. Financial Guarantees in Force Financial guarantee products were sold in three principal markets: the U.S. public finance market, the U.S. structured finance market and the international finance market. The following table provides a breakdown of guaranteed net par outstanding by market sector at September 30, 2012 and December 31, 2011. Guaranteed net par outstanding includes the exposures of policies that insure VIEs consolidated in accordance with ASC Topic 810, Consolidation: September 30, December 31, ($ in millions) 2012 2011 Public Finance $ 152,844 $ 176,817 Structured Finance 46,121 55,145 International Finance 39,144 40,542 Total net par outstanding $ 238,109 $ 272,504 Ratings Distribution The following tables provide a rating distribution of guaranteed total net par outstanding based upon internal Ambac Assurance credit ratings at September 30, 2012 and December 31, 2011 and a distribution by bond type of Ambac Assurance's below investment grade exposures at September 30, 2012 and December 31, 2011. Below investment grade is defined as those exposures with a credit rating below BBB-: 74 -------------------------------------------------------------------------------- Table of Contents Percentage of Guaranteed Portfolio(1) September 30, December 31, 2012 2011 AAA 1 % 1 % AA 22 24 A 44 43 BBB 18 17 Below investment grade 15 15 Total 100 % 100 % (1) Internal credit ratings are provided solely to indicate the underlying credit quality of guaranteed obligations based on the view of Ambac Assurance. In cases where Ambac Assurance has insured multiple tranches of an issue with varying internal ratings, or more than one obligation of an issuer with varying internal ratings, a weighted average rating is used. Ambac Assurance credit ratings are subject to revision at any time and do not constitute investment advice. Ambac Assurance, or one of its affiliates, has guaranteed the obligations listed and may also provide other products or services to the issuers of these obligations for which Ambac Assurance may have received premiums or fees. Summary of Below Investment Grade Exposure September 30, December 31, Bond Type 2012 2011 ($ in millions) Public Finance: Housing $ 775 $ 790 Tax-backed 742 764 Transportation 533 669 General obligation 499 492 Health care 51 79 Other 1,388 1,137 Total Public Finance 3,988 3,931 Structured Finance: Residential mortgage-backed and home equity-first-lien 10,120 11,673 Residential mortgage-backed and home equity-second-lien 7,847 8,784 Student loans 6,865 7,728 Structured Insurance 1,657 1,657 Mortgage-backed and home equity-other 450 586 Auto Rentals - 820 Enhanced equipment trust certificates - 401 Other 559 639 Total Structured Finance 27,498 32,288 International Finance: Airports 1,472 1,466 Other 3,453 3,399 Total International Finance 4,925 4,865 Total $ 36,411 $ 41,084 75 -------------------------------------------------------------------------------- Table of Contents The decrease in below investment grade exposures are primarily due to (i) reductions to residential mortgage-backed Securities during the year as result of prepayments by issuers and claims presented to Ambac Assurance; (ii) commutations of certain student loan policies during 2012 in addition to reductions to student loan policies during the year as a result of prepayments by issuers; (iii) prepayments of auto rental policies resulting in the full maturity of the related insurance policies; and (iv) an improvement in the credit quality of the enhanced equipment trust certificate policies resulting in their removal from below investment grade exposures. The sections that follow will discuss below investment grade exposures with residential mortgage-backed and student loan exposures. U.S. residential mortgage-backed securities exposure included in financial guarantee insurance portfolio The following tables provide current gross par outstanding by vintage and type, and underlying credit rating of Ambac's affected U.S. RMBS book of business: Total Gross Par Outstanding At September 30, 2012 ($ in millions) Year of Issue Second-Lien Sub-prime Mid-prime(1) Other(2) 1998-2001 $ 73.0 $ 598.0 $ 4.6 $ 476.1 2002 36.1 532.1 57.3 16.2 2003 26.7 808.8 382.1 170.2 2004 1,093.0 439.1 680.5 35.4 2005 1,112.0 1,030.2 2,087.8 60.0 2006 2,764.6 728.3 1,195.7 102.5 2007 2,979.0 487.9 2,079.1 266.5 Total $ 8,084.4 $ 4,624.4 $ 6,487.1 $ 1,126.9 % of Total RMBS Portfolio 39.8 % 22.8 % 31.9 % 5.5 % Gross claim liability, before Subrogation Recoveries At September 30, 2012 ($ in millions) Year of Issue Second-Lien Sub-prime Mid-prime(1) Other(2) 1998-2001 $ - $ 76.5 $ - $ - 2002 - 63.6 - - 2003 - 25.3 4.7 - 2004 112.3 35.6 41.0 - 2005 252.3 461.8 865.7 14.6 2006 1,481.7 465.4 854.0 119.4 2007 305.8 207.8 1,219.0 86.8 Total $ 2,152.1 $ 1,336.0 $ 2,984.4 $ 220.8 Gross Subrogation Recoveries At September 30, 2012 ($ in millions) Year of Issue Second-Lien Sub-prime Mid-prime(1) Other(2) 1998-2001 $ - $ - $ - $ - 2002 - - - - 2003 - - - - 2004 (43.6 ) - - - 2005 (138.8 ) (218.1 ) (246.9 ) - 2006 (797.0 ) (215.2 ) (171.7 ) - 2007 (566.2 ) (54.8 ) (236.6 ) - Total $ (1,545.6 ) $ (488.1 ) $ (655.2 ) $ - 76 -------------------------------------------------------------------------------- Table of Contents Percent of Related RMBS Transactions' Gross Par At September 30, 2012 Internal Ambac Credit Rating (3) Second-Lien Sub-prime Mid-prime(1) Other(2) AAA 0 % 4 % <1 % 12 % AA <0.1 % 3 % 4 % 24 % A 1 % 4 % 1 % 14 % BBB(4) 1 % 0 % 1 % 6 % Below investment grade(4) 98 % 89 % 94 % 44 % (1) Mid-prime includes Alt-A transactions and affordability product transactions, which includes interest only or option adjustable rate features. (2) Other includes primarily manufactured housing and lot loan exposures. (3) Ambac Assurance ratings set forth above reflect internal credit ratings as of September 30, 2012, and may be changed at any time based on our internal credit review. Ambac Assurance undertakes no obligation to update such ratings. This does not constitute investment advice. Ambac Assurance or one of its affiliates has guaranteed the obligations listed and may also provide other products or services to the issuers of these obligations for which Ambac Assurance may have received premiums or fees. (4) Ambac's BBB internal rating reflects bonds which are of medium grade credit quality with adequate capacity to pay interest and repay principal. Certain protective elements and margins may weaken under adverse economic condition and changing circumstances. These bonds are more likely than higher rated bonds to exhibit unreliable protection levels over all cycles. Ambac's below investment grade internal ratings reflect bonds which are of speculative grade credit quality with the adequacy of future margin levels for payment of interest and repayment of principal potentially adversely affected by major ongoing uncertainties or exposure to adverse conditions. Ambac Assurance's below investment grade category includes transactions on which claims have been submitted. Student Loans Our student loan portfolio consists of credits collateralized by (i) federally guaranteed loans under the Federal Family Education Loan Program ("FFELP") and (ii) private student loans. Whereas FFELP loans are guaranteed for a minimum of 97% of defaulted principal and interest, private loans have no government guarantee and, therefore, are subject to credit risk as with other types of unguaranteed credits. Private student loans are outside the purview of recent government programs designed to assist borrowers. Due to the failure of the auction rate and variable rate markets, the interest rates on student loan securities has increased significantly to punitive levels pursuant to the transaction terms. Such increases have caused the collateralization ratio in these transactions to deteriorate on an accelerated basis due to negative excess spread and/or the use of principal receipts to pay current interest. Effective July 1, 2010, lenders were unable to originate federal guaranteed loans, due to the termination of the FFELP program. The resulting reduction in new revenues may adversely affect a number of smaller issuers, whose ability to remain a going concern may be at risk. Student loan net par outstanding, excluding debt service reserve funds on Ambac insured obligations: September 30, December 31, 2012 % of Net 2011 % of Net Issuer Type ($ in millions) Net Par Par Net Par Par For-Profit Issuers $ 3,218.2 47 % $ 3,288.5 43 % Not-For-Profit Issuers 3,588.1 53 % 4,391.1 57 % Total $ 6,806.3 100 % $ 7,679.6 100 % Collateral for the For-Profit Issuers consists of private loans which do not have any federal guarantee as to defaulted principal and interest. Collateral for the Not-For-Profit Issuers consists of both FFELP and private loans. For the total student loan portfolio, approximately 73% of the collateral backing student loan trusts consists of private loans while the remaining 27% consists of FFELP loans. Private loan defaults have been on the rise since the credit crisis in 2008 began. Elevated unemployment rates, combined with high student loan debt levels will continue to put pressure on borrower's ability to pay their loans. The following table represents the student loan net par outstanding by underlying debt type, excluding debt service reserve funds on Ambac insured obligations: September 30, December 31, 2012 % of net 2011 % of net Debit Type ($ in millions) Net Par Par Net Par Par Auction Rate $ 4,123.6 60 % $ 4,644.6 61 % Fixed Rate 321.8 5 % 414.9 5 % VRDOs - - % 204.6 3 % Floating Rate Notes 2,360.9 35 % 2,415.5 31 % Total $ 6,806.3 100 % $ 7,679.6 100 % 77 -------------------------------------------------------------------------------- Table of Contents RESULTS OF OPERATIONS Ambac follows the accounting prescribed by ASC Topic 852, "Reorganizations". Entities operating in bankruptcy and expecting to reorganize under Chapter 11 of the Bankruptcy Code are subject to the additional accounting and financial reporting guidance in ASC Topic 852. While ASC Topic 852 provides specific guidance for certain matters, other portions of US GAAP continue to apply so long as the guidance does not conflict with ASC Topic 852. This accounting literature provides guidance for periods subsequent to a Chapter 11 filing for, among other things, the presentation of liabilities that are and are not subject to compromise by the Bankruptcy Court proceedings, as well as the treatment of interest expense and presentation of costs associated with the proceedings. Accordingly, the financial results in the prior periods or filed in future filings may not be comparable. Ambac's net income was $157.5 million or $0.52 per share, and a net loss of ($75.5) million or ($0.25) per share, for the three months ended September 30, 2012 and 2011, respectively. Ambac's losses attributable to common stockholders were ($400.3) million, or ($1.32) per share, and ($997.2) million or ($3.30) per share, for the nine months ended September 30, 2012 and 2011, respectively. The third quarter 2012 financial results compared to 2011 were positively impacted by (i) lower derivative product losses; (ii) a positive net change in fair value of credit derivatives; (iii) lower underwriting, interest, and reorganization expenses; and (iv) a lower provision for income taxes, partially offset by (i) lower income on variable interest entity activities and (ii) a lower benefit for loss and loss expenses. The nine months ended September 30, 2012 financial results were positively impacted by (i) a lower provision for loss and loss expenses; (ii) higher other income; (iii) lower derivative product losses; (iv) higher net realized investment gains; and (v) a lower provision for income taxes, partially offset by realized losses on extinguishment of surplus notes. The following paragraphs describe the consolidated results of operations of Ambac and its subsidiaries for the three and nine months ended September 30, 2012 and 2011 and its financial condition as of September 30, 2012 and December 31, 2011. Commutations, Terminations and Settlements of Financial Guarantee Contracts. A key business strategy for Ambac is to increase the residual value of its financial guarantee business by mitigating losses on poorly performing transactions via the pursuit of commutations and terminations of financial guarantee insurance contracts, including insurance policies and credit derivative contracts. For three and nine months ended September 30, 2012 and 2011, Ambac executed a number of such transactions as follows: In the first quarter of 2012, the Segregated Account of Ambac Assurance made a gross cash payment of $17.5 million ($10.5 million net of reinsurance) to commute $204.6 million of net par exposure relating to two student loan transactions. There were no payments or Surplus Notes issued during the second or third quarter of 2012 relating to commutations, terminations or settlements of financial guarantee or credit derivative transactions. In October 2012, the Segregated Account of Ambac Assurance made a gross payment of $38.0 million ($35.9 million net of reinsurance) to commute $325.3 million of net par exposure related to a student loan transaction. In the first quarter of 2011, the Segregated Account of Ambac Assurance completed the commutation of two student loan transactions representing $419.9 million of the net par exposure, with cash payment of $11.0 million and the issuance of Segregated Account Surplus Notes with a par value of $3.0 million. There were no payments or Surplus Notes issued during the second or third quarter of 2011 related to commutations, terminations or settlements of financial guarantee or credit derivative contracts. Net Premiums Earned. Net premiums earned for the three and nine months ended September 30, 2012 were $113.1 million and $311.1 million, respectively, an increase of $11.1 million, or 10.8%, from $102.0 million for the three months ended September 30, 2011 and an increase of $18.0 million, or 6.1%, from $293.1 million for the nine months ended September 30, 2011. Net premiums earned include accelerated premiums, which result from refunding, calls and other accelerations. Ambac had accelerated earnings of $34.4 million during the third quarter of 2012 versus $18.8 million in the third quarter of 2011 and $86.0 million for the nine months ended September 30, 2012 versus $30.1 million for the nine months ended September 30, 2011. The net increase in accelerated earnings was primarily driven by an increase in overall volume of calls of Ambac insured debt within the public finance market due to low interest rates and refinancings by healthcare providers. In addition, certain structured finance policies matured during the first nine months of 2011 with negative accelerations. Ambac recognizes negative accelerations on policies when the GAAP premiums receivable for the policy exceeds the policy's unearned premium reserve at termination. Normal net premiums earned, which exclude accelerated premiums for 2012, have been negatively impacted by the runoff of the insured portfolio either via transaction terminations, refundings or scheduled maturities. Premium receivables relating to a non-investment grade obligation were written-off as the premium was deemed uncollectable. For the nine months ended September 30, 2012, net premiums earned relating to this obligation deemed uncollectible and written off were negatively impacted by $13.8 million. Normal net premiums earned and accelerated premiums are reconciled to total net premiums earned in the table below and are included in the Financial Guarantee segment. The following table provides a breakdown of net premiums earned by market sector: 78 -------------------------------------------------------------------------------- Table of Contents Three Months Ended Nine Months Ended September 30, September 30, ($ in millions) 2012 2011 2012 2011 Public Finance $ 37.9 $ 40.3 $ 116.2 $ 125.3 Structured Finance 20.0 22.1 44.9 76.8 International Finance 20.8 20.8 64.0 60.9 Total normal premiums earned 78.7 83.2 225.1 263.0 Accelerated earnings 34.4 18.8 86.0 30.1 Total net premiums earned $ 113.1 $ 102.0 $ 311.1 $ 293.1 The following table provides a breakdown of accelerated earnings by market sector: Three Months Ended Nine Months Ended September 30, September 30, ($ in millions) 2012 2011 2012 2011 Public Finance $ 33.8 $ 18.3 $ 78.9 $ 40.6 Structured Finance 0.5 1.5 (4.3 ) (10.2 ) International Finance 0.1 (1.0 ) 11.4 (0.3 ) Total accelerated earnings $ 34.4 $ 18.8 $ 86.0 $ 30.1 Net Investment Income. Net investment income for the three and nine months ended September 30, 2012 was $84.1 million and $290.0 million, a decrease of 7% from $90.7 million for the three months ended September 30, 2011, and an increase of 8% from $267.8 million for the nine months ended September 30, 2011: Three Months Ended Nine Months Ended September 30, September 30, ($ in millions) 2012 2011 2012 2011 Financial Guarantee $ 81.0 $ 81.6 $ 276.2 $ 241.2 Financial Services 3.0 9.0 13.6 21.4 Corporate 0.1 0.1 0.2 0.2 Total net investment income $ 84.1 $ 90.7 $ 290.0 $ 262.8 The decrease in Financial Guarantee net investment income in the three months ended September 30, 2012 compared to the three months ended September 30, 2011, reflects the shift in portfolio holdings toward short-term securities in anticipation of the commencement of partial payment of Segregated Account claims, partially offset by the impact of a greater percentage of long-term holdings in higher yielding Ambac-insured securities. The increase in Financial Guarantee net investment income in nine months ended September 30, 2012, primarily reflects the benefit of holding a greater portion of the long-term portfolio in higher yielding Ambac-insured securities and a slightly higher average long-term invested asset base compared to the nine months ended September 30, 2011. The par value of Ambac Assurance-wrapped securities as of September 30, 2012 was $1,243 million compared to $1,076 million at September 30, 2011. The overall size of the Financial Guarantee long-term asset portfolio has declined by approximately $269 million since September 30, 2011 as continuing collection of installment paying financial guarantee premiums and coupon receipts on invested assets were offset by the resumption of partial claim payments on Segregated Account policies, payments made to commute certain financial guarantee exposures and the repurchase of surplus notes in the second quarter of 2012. The decrease in Financial Services investment income for the three and nine months ended September 30, 2012, compared to the three and nine months ended September 30, 2011, was driven primarily by the effects of a smaller portfolio of investments in the investment agreement business. The portfolio decreased as a result of sales of securities to fund repayment of intercompany loans and investment agreements as Ambac's investment agreement obligations were reduced from $590 million at September 30, 2011, to $416 million at September 30, 2012. Corporate investment income relates to the investments from Ambac's investment portfolio. Net Other-Than-Temporary Impairment Losses. Net other-than-temporary impairment losses recorded in Net income (loss) on the Consolidated Statements of Total Comprehensive Income include only the credit related impairment amounts to the extent management does not intend to sell and it is not more likely than not that the company will be required to sell before recovery of the amortized cost basis less any current period credit impairment. Non-credit related impairment amounts are recorded in Other 79-------------------------------------------------------------------------------- Table of Contents comprehensive income (loss). Alternatively, the non-credit related impairment would be recorded in Net other-than-temporary impairment losses within Net income (loss) on the Consolidated Statements of Total Comprehensive Income if management intends to sell the securities or it is more likely than not the company will be required to sell before recovery of amortized cost less any current period credit impairment. Charges for other-than-temporary impairment losses were $0.4 million and $5.7 million for the three and nine months ended September 30, 2012, respectively, and $10.9 million and $30.2 million for the three and nine months ended September 30, 2011. As further described in Note 1 to the Unaudited Consolidated Financial Statements located in Part I, Item 1 of this Form 10-Q, on March 24, 2010, OCI commenced the Segregated Account Rehabilitation Proceedings in order to facilitate an orderly run-off and/or settlement of the liabilities allocated to the Segregated Account. As a result of actions taken by the OCI, financial guarantee payments on securities guaranteed by Ambac Assurance which have been allocated to the Segregated Account are no longer under the control of Ambac management. Claim payments under Segregated Account policies remained suspended throughout 2011 and into the third quarter of 2012. On September 20, 2012, in accordance with published Policy Claim Rules, the Segregated Account commenced paying 25% of each permitted policy claim that arose since the commencement of the Segregated Account Rehabilitation Proceedings. The Segregated Account will continue to make cash payments of 25% of each policy claim submitted and permitted in accordance with such Policy Claim Rules, subject to any further orders of the Rehabilitation Court. As further described in Note 1 to the Unaudited Financial Statements located in Part 1 Item 1 of this Form 10-Q, no decision has been announced with respect to effectuating or amending the Segregated Account Rehabilitation Plan. Possible modifications to the Plan, with respect to the form and timing of satisfying the remaining balance of unpaid claims, could have a material effect on the recognition of other-than-temporary impairment for Ambac-wrapped securities. The Rehabilitator has previously announced that more specific information regarding the status of the Segregated Account Rehabilitation Plan, including possible modifications, will be provided as soon as appropriate. Since commencement of the Segregated Account Rehabilitation Proceedings, changes in the estimated timing of claim payments have resulted in adverse changes in projected cash flows on certain impaired Ambac-wrapped securities. Such changes in estimated claim payments on Ambac-wrapped securities contributed to net other-than-temporary impairments for the three and nine months ended September 30, 2012 and 2011. Additionally, the nine month period ended September 30, 2012 and the three and nine months ended September 30, 2011 included credit impairments on certain other non-agency RMBS securities. As of September 30, 2012, management has not asserted an intent to sell any securities from its portfolio that are in an unrealized loss position. Future changes in our estimated liquidity needs could result in a determination that Ambac no longer has the ability to hold such securities, which could result in additional other-than-temporary impairment charges. Net Realized Investment Gains. The following table provides a breakdown of net realized gains for the three and nine months ended September 30, 2012 and 2011: Financial Financial ($ in millions) Guarantee Services Corporate Total Three months ended September 30, 2012: Net gains on securities sold or called $ 3.4 $ - $ - $ 3.4 Foreign exchange losses (0.2 ) - - (0.2 ) Total net realized gains $ 3.2 $ - $ - $ 3.2 Three months ended September 30, 2011: Net gains on securities sold or called $ 3.2 $ 0.4 $ - $ 3.6 Foreign exchange gains 1.5 - - 1.5 Total net realized gains $ 4.7 $ 0.4 $ - $ 5.1 Financial Financial ($ in millions) Guarantee Services Corporate Total Nine months ended September 30, 2012: Net gains on securities sold or called $ 43.9 $ 27.1 $ - $ 71.0 Foreign exchange losses (0.4 ) - - (0.4 ) Total net realized gains $ 43.5 $ 27.1 $ - $ 70.6 Nine months ended September 30, 2011: Net gains on securities sold or called $ 0.8 $ 2.7 $ - $ 3.5 Foreign exchange gains 1.5 - - 1.5 Total net realized gains $ 2.3 $ 2.7 $ - $ 5.0 80 -------------------------------------------------------------------------------- Table of Contents Change in Fair Value of Credit Derivatives. The net change in fair value of credit derivatives resulted in gains of $27.4 million and $12.8 million for the three and nine months ended September 30, 2012, respectively, compared to gains of $4.5 million and $19.9 million in the three and nine months ended September 30, 2011, respectively. The net gain on change in fair value of credit derivatives for the three month period ended September 30, 2012 resulted from improvement in reference obligation prices, gains associated with runoff of the portfolio and CDS fees earned. The net gain on change in fair value of credit derivatives for the three months ended September 30, 2011 resulted primarily from CDS fees earned and the reversal of unrealized losses associated with terminations, partially offset by declines in certain reference obligation prices particularly related to student loan securitizations. There was no change to the Ambac credit valuation adjustment ("CVA") to reflect Ambac's own credit risk in the fair value of credit derivatives during the three month periods ended September 30, 2012 and 2011. The net gain on change in fair value of credit derivatives for the nine month periods ended September 30, 2012 and 2011 resulted from improvement in reference obligation prices, gains associated with runoff of the portfolio and CDS fees earned, partially offset by reductions of the Ambac CVA. Reductions of the Ambac CVA in these periods reflect increases in the market value of other securities containing significant Ambac credit risk. Refer to Note 7 to the Unaudited Consolidated Financial Statements in Part I, Item 1of this Form 10-Q for further information about the determination of fair value of credit derivatives, including the Ambac CVA. Of the net change in fair value of credit derivatives reported for each period, reductions to the Ambac CVA account for losses of $0 million and $0 for the three month periods ended September 30, 2012 and 2011, respectively; and losses of $63.8 million and $47.2 million for the nine month periods ended September 30, 2012 and 2011, respectively. Realized gains and other settlements on credit derivative contracts were $2.9 million and $9.3 million for the three and nine months ended September 30, 2012, respectively; and $3.8 million and $13.4 million for the three and nine months ended September 30, 2011, respectively. These amounts represent premiums received and accrued on written contracts. The declines period to period are due to reduced premium receipts resulting from runoff of the credit derivative portfolio. There were no loss or settlement payments in the periods presented. Unrealized gains on credit derivative contracts were $24.5 million and $3.5 million in the three and nine months ended September 30, 2012, respectively, compared to $0.7 million and $6.5 million for the three and nine months ended September 30, 2011, respectively. The net unrealized gains (losses) in fair value of credit derivatives reflect the same factors as the overall change in fair value of credit derivatives as noted above excluding the impact of premium revenues reported within Realized gains and other settlements. Derivative Product Revenues. The net losses reported in Derivative product revenues were $36.0 million and $113.2 million for the three and nine months ended September 30, 2012, respectively, compared to $215.8 million and $260.4 million for the three and nine months ended September 30, 2011, respectively. The net loss for the three months ended September 30, 2012 primarily resulted from a realized loss related to the negotiated termination of a derivative asset. Interest rate movements did not have a significant impact on results for the quarter. For the nine months ended September 30, 2012 and the three and nine month periods ended September 30, 2011 derivative product losses resulted primarily from mark-to-market movements in the portfolio caused by declining interest rates in each of the periods. The derivative products portfolio is positioned to benefit from rising rates as an economic hedge against interest rate exposure in the financial guarantee portfolio (the "macro-hedge"). This additional interest rate sensitivity contributed losses of ($1.0) million and ($53.8) million for the three and nine months ended September 30, 2012, respectively; compared to ($203.5) million and ($262.9) million for the three and nine months ended September 30, 2011, respectively. Excluding the impact of the macro-hedge, derivative products revenue resulted in (losses) gains of ($35.1) million and ($59.4) million for the three and nine months ended September 30, 2012, respectively; compared to ($12.3) million and $2.5 million for the three and nine months ended September 30, 2011, respectively. Losses excluding the macro-hedge for the three and nine months ended September 30, 2012 primarily resulted from the negotiated termination discussed above and decreases in the Ambac CVA on certain customer swap liabilities discussed below. Additionally, the nine months ended September 30, 2012 included losses arising from the impact of changes in interest rates on the fair value of certain customer swaps. For the three and nine months ended September 30, 2011, customer swaps experienced mark-to-market losses from interest rates offset by positive movements in projected inflation and were also impacted by the CVA as described below. The fair value of derivatives includes a valuation adjustment to reflect Ambac's own credit risk when appropriate (Ambac "CVA"). Within the financial services derivatives portfolio, an Ambac CVA is generally applicable for uncollateralized derivative liabilities that may not be offset by derivative assets under a master netting agreement. Inclusion of an Ambac CVA in the valuation of financial services derivatives resulted in gains (losses) within derivative products revenues of ($5.3) million and ($11.6) million for the three and nine months ended September 30, 2012, respectively; compared to $92.4 million and $92.9 million for the three and nine months ended September 30, 2011, respectively. 81 -------------------------------------------------------------------------------- Table of Contents Net Realized (Losses) Gains on Extinguishment of Debt. During June 2012, Ambac Assurance exercised options to repurchase surplus notes with an aggregate par value of $789.2 million for an aggregate cash payment of $188.4 million. These surplus notes were issued in connection with the settlement of credit derivative liabilities in 2010 and were recorded at their fair value at the date of issuance. This fair value was at a significant discount to par value and since 2010 Ambac has accreted the discount into earnings using the effective interest method. As described further under "Other (Loss) Income" below, certain of these options were free-standing derivatives for accounting purposes and were carried at fair value as assets on the Consolidated Balance Sheets. The carrying value of extinguished surplus notes and accrued interest less the fair value of the free-standing derivatives were below the call option exercise prices and, accordingly, for the nine months ended September 30, 2012, Ambac recognized a realized loss of $177.7 million. Refer to Note 10 to the Unaudited Consolidated Financial Statements in Part I, Item 1 of this Form 10-Q for further information on the repurchase of surplus notes. There were no realized gains (losses) from settlements of investment agreements in the three and nine month periods ended September 30, 2012, or in the three month period ended September 30, 2011. Net realized (losses) gains on extinguishment of debt included gains from the settlement of investment agreements below their carrying value of $3.1 million for the nine month period ended September 30, 2011. Other (Loss) Income. Other (loss) income for the three and nine months ended September 30, 2012 was ($0.4) million and $100.6 million, respectively, compared to ($5.0) million and $32.6 million for the three and nine months ended September 30, 2011, respectively. Included within Other (loss) income are mark-to-market gains and losses relating to options to call certain Ambac Assurance surplus notes, non-investment portfolio related foreign exchange gains and losses, deal structuring fees, commitment fees and reinsurance settlement gains (losses). Other (loss) income for the nine months ended September 30, 2012 primarily resulted from mark-to-market gains of $100.7 million, related to Ambac's option to call $500 million par of bank surplus notes which were free-standing derivatives for accounting purposes. Such options were exercised in June 2012. Changes in fair value of these options through the date of exercise are reported within Other (loss) income. Other (loss) income for the three and nine months ended September 30, 2011 includes mark-to-market (losses) gains of ($6.9) million and $13.9 million, respectively, related to the surplus note call options. Income on variable interest entities. Income on variable interest entities for the three and nine months ended September 30, 2012 was $6.1 million and $26.9 million, respectively, compared to $55.0 million and $51.2 million for the three and nine months ended September 30, 2011, respectively. Included within Income on variable interest entities are amounts relating to VIEs consolidated under ASC Topic 810, including gains or losses attributable to consolidating or deconsolidating VIEs during the period reported. Generally, the Company's consolidated VIEs are entities for which Ambac has provided financial guarantees on its assets or liabilities. In consolidation, most assets and liabilities of the VIEs are reported at fair value and the related insurance assets and liabilities are eliminated. Differences between the net carrying value of the insurance accounts and the carrying value of the consolidated VIE's net assets are recorded through income at the time of consolidation or deconsolidation. From April 1, 2011 through December 31, 2011, the Company consolidated one VIE that contained primarily intangible assets associated with the operations of the VIE's subsidiaries. These intangible assets were amortized through earnings over time and were subject to potential impairment. The VIE's note liabilities were carried at fair value with changes in value reported through earnings. In consolidation, adverse changes in the estimated cash flow available to pay the note liabilities of this insured transaction reflect increased credit risk that will generally result in declines to the fair value of those liabilities and corresponding mark-to-market gains recognized through income. The gains reported in Income on variable interest entities for the three and nine months ended September 30, 2012 reflect positive changes in the fair value of net assets of consolidated VIEs during those periods. The net gains for the three and nine months ended September 30, 2011 primarily related to the VIE consolidated effective April 1, 2011 and containing intangible assets described above. For the three months ended September 30, 2011, adverse changes in the expected performance of this transaction resulted in a net gain of $53.1 million, including a $103.0 million reduction to the fair value of the VIE note liabilities and other income, partially offset by an impairment loss to the VIE's intangible assets. Income on variable interest entities for the nine months ended September 30, 2011 included a $55.5 million gain from the April 1, 2011 consolidation of this VIE, mark-to-market gains of $30.2 million on the change in fair value of the related note liabilities, and other operating gains, partially offset by the impairment charge on intangible assets described above. The nine month period ended September 30, 2011 also includes positive changes in the fair value of net assets of other consolidated VIEs during the period. Losses and Loss Expenses. Losses and loss expenses are based upon estimates of the aggregate losses inherent in the non-derivative financial guarantee portfolio for insurance policies issued to beneficiaries, including VIEs, for which we do not consolidate the VIE. Loss and loss expenses for the three and nine months ended September 30, 2012 were $(18.7) million and $720.3 million, respectively, compared to ($60.2) million and $1,055.8 million the three and nine months ended September 30, 2011, respectively. Losses for the three months ended September 30, 2012 were driven by lower estimated losses in first lien and second lien RMBS credits, partially, offset by an increase in estimated losses for certain student loan transactions and asset-backed credits. Losses for the nine months ended September 30, 2012 were driven by higher estimated losses in the first-lien and second-lien RMBS portfolio and certain structured insurance and asset-backed credits. Losses for the three months ended September 30, 2011 were driven by lower estimated losses in the first-lien RMBS and student loan portfolios, offset by an increase in estimated losses for the second-lien RMBS portfolio and for certain structured finance insurance policies. Losses for the nine months ended September 30, 2011 were driven by higher estimated losses in the first-lien RMBS and student loan portfolios, offset by a decrease in estimated losses for the second-lien RMBS portfolio. 82-------------------------------------------------------------------------------- Table of Contents The following table summarizes the changes in the total net loss and loss expense reserves for the nine months ended September 30, 2012 and the year ended December 31, 2011: Nine Months Ended Year Ended ($ in millions) September 30, 2012 December 31, 2011 Beginning balance of loss and loss expense reserves, net of subrogation recoverable and reinsurance $ 6,230.8 $ 4,424.5 Provision for losses and loss expenses 720.3 1,859.5 Paid claims and loss expenses (769.3 ) (308.6 ) Recoveries of losses and loss expense paid from reinsurers 14.9 21.0 Other recoveries, net of reinsurance 135.3 105.7 Other adjustments (including foreign exchange) 21.9 (5.9 ) Net deconsolidation of certain VIEs - 134.6 Ending balance of net loss and loss expense reserves, net of subrogation recoverable and reinsurance $ 6,353.9 $ 6,230.8 Refer to Note 5 to the Unaudited Consolidated Financial Statements in Part I, Item 1 of the Form 10-Q for further information on the basis for consolidations and deconsolidations of VIEs. The losses and loss expense reserves as of September 30, 2012 and December 31, 2011 are net of estimated recoveries under representation and warranty breaches for certain RMBS transactions in the amount of $2,688.9 million and $2,720.3 million, respectively. Please refer to the "Critical Accounting Estimates" section of this Management's Discussion and Analysis of Financial Condition and Results of Operations and to Note 6 of the Unaudited Consolidated Financial Statements in Part I, Item 1 of this Form 10-Q for further background information on the change in estimated recoveries. The following tables provide details of net reported claims, net of recoveries received, for the nine months ended September 30, 2012 and 2011: |